GPT-6.1 Sol Ultrafast ships: up to 8x faster coding in Codex
The speed tier teased at DevDay is rolling out — at a steep premium, Pro 500 only, with EU data residency and a plain-English test agent landing too.
Copy markdown
8x faster loops in Codex
OpenAI's Ultrafast tier for GPT-6.1 Sol — the "coming soon" option from DevDay — has reportedly gone live, pushing up to 8x faster token generation in Codex (~300 tokens/sec) and up to 6x in the API. For an edit-run-debug loop, that's the gap between waiting on the model and barely noticing it.
The catch: a steep speed premium
Ultrafast reportedly costs about 6x the standard rate (Sol lists at $2 / $10 per million input/output tokens), and it's gated to Pro 500 and eligible Enterprise/Edu plans. Great for latency-critical agents; run the math before flipping it on for a high-volume job.
EU data residency reaches the Fast tiers
Alongside the rollout, GPT-6.1 Sol Fast and GPT-6 Luna Fast reportedly gained EU data residency. EU-based teams can now put the fast tiers to work without routing requests outside the region.
Functionize ships an MCP server for coding agents
Ask Claude Code, Cursor, or Copilot in plain English for a smoke or regression suite and Functionize Studio builds it, runs it against your app, and reports the first failing step back in the same chat. Works with any MCP-compatible client and is open to Studio customers and trial users today.